Agriculture, Ecosystems and Environment 193 (2014) 60–69

Contents lists available at ScienceDirect

Agriculture, Ecosystems and Environment

journal homepage: www.elsevier.com/locate/agee

Agrochemicals in field margins – An experimental field study to assess

the impacts of pesticides and fertilizers on a natural plant community
Juliane Schmitz ∗ , Melanie Hahn, Carsten A. Brühl
Institute for Environmental Sciences, University Koblenz-Landau, Fortstraße 7, 76829 Landau, Germany

a r t i c l e i n f o a b s t r a c t

Article history: In agricultural areas, field margins are often the only remaining habitat for wild plant species. However,
Received 19 December 2013 due to their proximity to agricultural fields, the vegetation of field margins may be affected by agro-
Received in revised form 15 April 2014 chemicals applied to the crop field. To investigate individual and combined effects of fertilizer, herbicide,
Accepted 25 April 2014
and insecticide inputs on the plant community of field margins, a 3-year field study with a randomized
Available online 20 May 2014
block design was performed. The applied fertilizer rates (25% of the field rate) and pesticide rates (30%
of the field rate) were consistent with their average input rates (drift + overspray) in the first meter of a
Keywords:
field margin directly adjacent to the field. Fertilizer and herbicide applications resulted in significantly
Agro-ecosystem
Off-field habitats reduced frequencies of several plant species. The fertilizer promoted plants with a high nutrient uptake
Non-target species and decreased the frequencies of small and subordinate species. In addition to the disappearance of a few
Plant frequency species, the herbicide caused predominantly sublethal effects, which gradually reduced the frequencies
Interaction effects of certain species. Significant herbicide–fertilizer interaction effects were also observed and could not be
Plant diversity extrapolated from individual effects. The impacts of both agrochemicals became stronger over time, led
to shifts in plant community compositions, and caused significantly lower species diversities than in the
control plots. The insecticide application significantly affected the frequencies of two plant species. The
results suggest that a continuous annual application of agrochemicals would cause further plant com-
munity shifts. Hence, to preserve biodiversity of agricultural landscapes, it is recommended to protect
the vegetation in field margins from agrochemical inputs.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction graphical information systems indicated that field margins with a

A decrease in biological diversity in farmlands across Europe
has been observed over the last several decades, and agricul-
tural intensification was identified as a major driving force of this
decrease (Berendse et al., 2004; Tscharntke et al., 2005). Agricul-
tural intensification caused land-use changes, such as increases in
farm size, specialization, and management intensity (Tscharntke
et al., 2005; Firbank et al., 2008). Consequently, complex natural
ecosystems and semi-natural habitats have decreased in number
and size (Benton et al., 2003). Today, the majority of semi-natural
habitats in agricultural landscapes are field margins (Marshall and
Moonen, 2002). These landscape structures are usually linear, per-
manent vegetation strips of grassy and herbaceous off-field habitats
adjacent to agricultural fields (Kühne and Freier, 2001; Kleijn and
Verbeek, 2000; Hahn et al., 2014). Generally, these habitats are
only a few meters wide. A study using digital orthophotos and geo-
∗ Corresponding author. Tel.: +49 (0) 6341 280 31319; fax: +49 (0) 6341 280 31326.
E-mail address: Schmitz@uni-landau.de (J. Schmitz).
width of 1–2 m are the typical margins remaining in intensively
used agricultural landscapes in Germany (Hahn et al., 2014).
Field margins are beneficial for the conservation of biodiver-
sity because they are often the only remaining habitat of a variety
of wild plant species and farmland animals in agro-ecosystems
(Nentwig, 2000; Asteraki et al., 2004 and references therein).
However, the biodiversity of field margins can be affected by
agrochemicals due to the proximity of these habitats to agricul-
tural fields (Firbank et al., 2008). For example, the vegetation of
field margins may be exposed to herbicides. Studies have shown
that herbicide applications in field margins can negatively affect the
plant community composition (Kleijn and Snoeijing, 1997; Marrs
and Frost, 1997; de Snoo and van der Poll, 1999). In Germany in par-
ticular, narrow field margins appear to be strongly affected. This is
because field margins that are less than 3 m wide are not considered
as terrestrial non-target areas and therefore, are not protected from
herbicide inputs by risk mitigation measures (e.g., in-field buffer
zone distances to terrestrial non-target areas) (Kühne et al., 2000;
BVL, 2013). Consequently, farmers in Germany and in other Euro-
pean countries do not have to maintain distances from field margins

http://dx.doi.org/10.1016/j.agee.2014.04.025
0167-8809/© 2014 Elsevier B.V. All rights reserved.
 J. Schmitz et al. / Agriculture, Ecosystems and Environment 193 (2014) 60–69
during pesticide applications and thus, these field margins receive
pesticide inputs via overspray and spray drift (Schmitz et al., 2013).
Fertilizer misplacements in field margins are also supposed to
affect the vegetation of field margins. Fertilizers are usually applied
on the field using spreaders, which distribute the fertilizer via spin-
ning disks that eject fertilizer backwards and sideways from the
spreader. Thus, fertilizer misplacements in field margins are likely
to occur (Rew et al., 1995; Tsiouris and Marshall, 1998; Wilson,
1999).
Furthermore, because fertilizers and herbicides are both
designed to affect vegetation, their application to field margins
will most likely involve interactions with each other. However, to
date, only a few studies have been concerned with such combined
effects on natural plant communities (Perry et al., 1996; Kleijn and
Snoeijing, 1997; Gove et al., 2007; Strandberg et al., 2012). These
studies demonstrated that herbicide and fertilizer inputs below
recommended crop application rates can significantly affect the
plant community.
Another concern involves the annual application sequences of
agrochemicals on a field. A conventionally managed winter wheat
field is treated annually with fertilizer, at least once with an herbi-
cide, and also with an insecticide. Insecticides could probably cause
indirect effects on plants by reducing herbivorous or flower-visiting
insects, which have been less investigated until now. Furthermore,
the annual repeated exposure of the vegetation to agrochem-
icals might intensify these effects and/or cause cumulative
effects.
The aim of this study was to investigate effects of agrochem-
ical misplacements (fertilizer, herbicide, and insecticide) on the
plant community of a field margin during three successive grow-
ing seasons (2010–2012). The study was specifically designed
to separate the effects of these three stressors from each other
and to investigate their combined effects because a field mar-
gin of a conventionally arable field is exposed to all of these
stressors.
2. Materials and methods
2.1. Study site
The study site was an extensively managed hay meadow
that was mowed twice per year without any fertilizer or pes-
ticide applications. The meadow was 1 ha in size, located near
Landau (South Rhineland Palatinate, Germany), and consisted
of a semi-natural species-rich plant community (belonging to
the Molinio-Arrhenatheretea meadows, Ellenberg et al., 1992)
containing 54 species (40 herbs, 14 grasses, based on vegeta-
tion assessments conducted in May and June). All species and
supplementary information (life span, type of reproduction, Ellen-
berg‘s indicator value for nitrogen, German Red List status) are
listed in Appendix A. The overall natural distribution of plant
species was homogenous across the meadow. A few of the species
were naturally more abundant than others, and consequently
these species were found more frequently (26 species; a species
was classified as common when at least two individuals per
plot were documented; see chapter 2.3 for details of vegetation
assessments).
The field experiment was established in spring 2010 and was
designed to study individual and combined effects of repeated
agrochemical applications on a surrogate field margin in succes-
sive growing seasons (Schmitz et al., 2013). We used a randomized
block design with seven treatments and one control. The treat-
ments consisted of three single applications, i.e., – one fertilizer (F),
one herbicide (H), and one insecticide (I); and all possible combina-
tions of these treatments (F + I, H + I, F + H, F + H + I); and one control
61
(C). Hence, the experiment was planned as a fully factorial design
where the three factors (fertilizer, herbicide, and insecticide) had 2
levels (applied versus not applied) resulting in a 2 × 2 × 2 factorial
design. All of the treatments (including the control) were repli-
cated eight times in plots, resulting in a total of 64 plots. Each plot
measured 8 m × 8 m, and 2 m of distance separated adjacent plots
(Schmitz et al., 2013).
2.2. Agrochemical applications
The majority of farmed fields in Germany are winter wheat
fields, and the selected meadow served as a surrogate for field
margins adjacent to such fields. Therefore, the field management
of winter wheat fields, with their agrochemical applications and
application sequences, was imitated. Fertilizer and pesticide rates
used for the plot applications were equal to their average input
rates in the first meter of a field margin directly adjacent to such a
field.
During fertilizer application, there is usually an input rate of
25% of the field rate in the first meter of a field margin (Tsiouris
and Marshall, 1998), and during pesticide application, the average
input rate in the first meter of a field margin is approximately 30%
of the field rate (direct overspray and spray drift) (see Schmitz et al.,
2013 for details). The recommended field rate of fertilizer is 200 kg
nitrogen (N)/ha per year, which should be applied in two equal rates
(100 kg N/ha each), one at the beginning of the vegetation period in
spring (when the wheat starts to grow) and the second a few weeks
later (personal communications with farmers and agrochemical
suppliers). We applied a granular NPK (nitrate, phosphorus, potas-
sium) fertilizer (14% N, Floral Düngemittel) and a calcium carbonate
and ammonium nitrate fertilizer (27% N; Raiffeisen Markt) at the
beginning of April (NPK fertilizer) and approximately 3 weeks
later (calcium carbonate and ammonium nitrate fertilizer) in 2010,
2011, and 2012. Each time, 25 kg N/ha (=25% of the field rate) was
applied. The fertilizer was applied using a hand-operated fertilizer
spreader (Power Spreader by Wolf Garten; MTD Products Aktienge-
sellschaft).
We used the herbicide Atlantis WG (sulfonylurea; recom-
mended field rate 400 g/ha, active ingredients [a.i.] 30 g/kg
mesosulfuron-methyl, 6 g/kg iodosulfuron-methyl-natrium,
90 g/kg mefenpyr-diethyl [Safener], mode of action: inhibitor of
plant cell division [e.g., acetolactate synthase], Bayer CropScience).
We applied the product once per year in April 2010, 2011, and
2012. The insecticide Karate Zeon (pyrethroid; recommended
field rate 75 ml/ha, a.i. lambda-cyhalothrin 7.5 ml a.i./ha, mode of
action: nonsystemic insecticide with contact and stomach action,
repellent properties, Syngenta) was applied once per year at the
end of May or at the beginning of June 2010, 2011, and 2012. These
pesticides were chosen because they were among the five most
commonly used pesticides in winter wheat fields in Germany at
the beginning of the study (Freier et al., 2008). The application rates
of the herbicide and the insecticide were 120 g Atlantis WG/ha
and 22.5 ml Karate Zeon/ha, respectively, which represented 30%
of their corresponding recommended field application rates. Each
time, the plots were treated under good agricultural practice (wind
speed <5 m/s, temperature <25 ◦ C, no rain 1 day before and after
application). The products were applied using a purpose-built
and air-assisted experimental field sprayer on wheels (Schachtner
Gerätetechnik). The field sprayer was equipped with an 8-m spray
boom with 15 flat-fan TeeJet nozzles (XR 11002-VS; Schachtner
Gerätetechnik). The boom height above the vegetation canopy
and distance between the nozzles was 50 cm. Following label
recommendations for field applications, a spray volume of 400 l/ha
was used.
62 J. Schmitz et al. / Agriculture, Ecosystems and Environment 193 (2014) 60–69
2.3. Plant community assessments
The plant community was assessed using the frequency method
(Elzinga et al., 1998) with a mapping frame of 1 square meter, which
was divided into 25 sub-squares of 20 cm × 20 cm. We placed the
frame on top of the vegetation and recorded the presence of each
plant species in each sub-square. This method allows for the iden-
tification of changes in plant communities over time (Elzinga et al.,
1998). An additional advantage of this method is that a uniform
plant community assessment can be obtained because the only
decision required by the observer is whether the species is present
within the sub-square (Elzinga et al., 1998). Study staff was compre-
hensively trained in species identifications. The plant community
assessments were conducted six times per plot along the diago-
nal of the plots. All of the community assessments (6 vegetation
assessments per plot × 64 plots = 384) were performed within a 1-
week period in mid-June of each year (2010, 2011, and 2012). We
calculated the frequency of each species per plot and treatment.
If a species was recorded 150 times per plot (25 sub-squares × 6
assessments per plot), then this species exhibited a plant frequency
of 100%.
2.4. Biomass samples
Plant biomass samples were collected at the end of June in 2010,
2011, and 2012 by cutting the above-ground plant biomass in one
quadrant measuring 1 m × 1 m in the middle of each plot. The fresh
weights of the samples were recorded in the field immediately after
cutting.
2.5. Statistical analyses
2.5.1. Univariate analysis
The data from each year (2010, 2011, and 2012) were ana-
lyzed separately. The statistic program Primer (Version 6) with the
Permanova+ add-on was used (Anderson et al., 2008). We tested
the differences in the number of species, species diversities, and
biomass samples between the seven treatments and the control
with a one-way permutational univariate analysis of variance (Per-
Anova). Euclidean distance was used as a distance measurement
to generate resemblance matrices. To test at a significance level of
0.001, 9999 permutations were generated (Anderson et al., 2008).
Post-hoc PerAnova pairwise comparisons were adjusted using a
Bonferroni correction.
To test for treatment effects on individual species, a nested Per-
Anova was performed. We used a mixed-effect model design with
the block as a random factor and the treatment as a fixed factor.
The six vegetation assessments per plot were nested within the
factor treatment (settings: Euclidean distance, 9999 permutations,
and Bonferroni corrections).
In addition, the factorial design of the experiment allowed us to
test for main effects of individual factors (F, H, and I) and interaction
effects between the three factors (F × H, F × I, H × I, F × H × I). The
main effect is the overall (or average) effect of one factor averaged
across the levels of other factors. An interaction effect occurs when
the effect of one factor depends on another factor. We performed a
three-factorial PerAnova to analyze main and interaction effects on
the number of species, biomass, and plant frequency of individual
species. The analyses were performed with the Euclidean distance
and 9999 permutations.
2.5.2. Multivariate analysis
Nonmetric multidimensional scaling (NMDS) was performed to
examine similarities in plant community compositions among the
seven treatments and the control. The data from each year (2010,
2011, and 2012) were analyzed separately. Statistical analyses were
performed using the open-source software R (www.r-project.org,
version 3.01) and the Biodiversity R package (Kindt and Coe, 2005).
We performed a two-dimensional NMDS based on the species
data (obtained from plant community assessments in June) in each
plot. Those species with a plant frequency of <1% were excluded.
The ordination method NMS standard (function NMSrandom), the
Bray–Curtis dissimilarity index, and 100 NMS permutations were
used.
In addition, we performed a two-dimensional NMDS with
weighted-average species scores and a subsequent vector fitting
of environmental variables (function metaMDS in the R package
vegan (Oksanen, 2013)) with the data from the plant commu-
nity assessments in June 2012. This analysis were performed to
present individual species, individual factors (F, H, I) and the con-
trol (C) on one ordination plot and to check whether certain species
are associated with specific factors (or the control) after the third
experimental season. We used the Bray–Curtis dissimilarity index,
and the NMDS was started with a maximum of 50 random starts.
Species scores and significant vectors (999 permutations) were
added to the ordination plot (function envfit in the R package vegan
(Oksanen, 2013)).
Permutational multivariate analysis of variance (PerManova)
was additionally performed to find significant differences in species
compositions among the treatments and the control. The data from
each year (2010, 2011, and 2012) were analyzed separately. We
used the Bray–Curtis dissimilarity index, 9999 permutations and
Bonferroni corrections. The analyses were performed using Primer
with the Permanova+ add-on (Anderson et al., 2008).
3. Results
3.1. Number of species
The mean number of species per treatment changed only slightly
over the years (2010–2012). We observed a significantly lower
mean number of dicotyledons in the treatment combinations of
fertilizer and herbicide (F + H, and F + H + I) than in the control plots
during the third experimental season (Fig. 1). The mean number
of monocotyledons (grasses) was not significantly affected by the
treatments during the experiment.
Analysis of main effects (three factorial PerAnova) indicated a
significant main effect of fertilizer on the mean number of dicotyle-
dons in June 2011 (p = 0.007) and in June 2012 (p = 0.002). In
addition, a significant main effect of the factor herbicide (p = 0.03)
was found in June 2012. Interaction effects between factors were
not identified.
3.2. Mean plant frequency of individual species
The mean frequencies of most species gradually changed in the
plots treated with fertilizer and herbicide during the experiment.
For example, in 2010, the treatment combination of fertilizer, herbi-
cide, and insecticide (F + H + I) significantly altered the frequencies
of nine species (Galium mollugo, Lathyrus pratensis, Vicia sepium,
Rhinanthus alectorolophus, Glechoma hederacea, Stellaria graminea,
Vicia hirsuta, Trifolium pratense, and Heracleum sphondylium (Table
S2 in Appendix A). Over time, significant treatment effects became
more apparent. As a result, the number of treatment effects, e.g., in
the F + H + I treatment, nearly doubled in third year compared with
the first year (Tables 1 and 2).
Based on the results listed in Tables 1 and 2, three primary
groups of responses of individual species to the treatments were
recognized after 3 years, which are described in the following sec-
tions and summarized in Table 3.
 J. Schmitz et al. / Agriculture, Ecosystems and Environment 193 (2014) 60–69 63

Fig. 1. Mean number (±standard error) of dicotyledons and monocotyledons in June 2010, 2011, and 2012; replicates per treatment = 8. * Significantly different from the
control, p < 0.05 (one-way analysis of variance with permutations [PerAnova]); p values Bonferroni corrected. C = control, I = insecticide, F = fertilizer, H = herbicide.

3.2.1. Response 1: 3.2.2. Response 2:

Eight species (G. hederacea, Ranunculus acris, L. pratensis, V.
sepium, Rumex acetosa, V. hirsuta, R. alectorolophus, and Anthox-
anthum odoratum) were significantly negatively affected by the
separate herbicide and fertilizer treatment but showed no sig-
nificant response to the insecticide treatment. In addition, we
detected significant herbicide–fertilizer interaction effects on
three of these species: L. pratensis (factorial PerAnova: F × H:
p = 0.007), R. alectorolophus (factorial PerAnova: F × H: p = 0.03), and
A. odoratum (factorial PerAnova: F × H: p = 0.001) (Table 1, last col-
umn).
Six species (Ajuga reptans, Veronica chamaedrys, Plantago lance-
olata, Lotus corniculatus, Agrostis capillaris, and Holcus lanatus) were
significantly negatively affected by the separate fertilizer treat-
ment but not by the herbicide or insecticide treatment. However, it
appears that the herbicide in the combined fertilizer and herbicide
treatments (F + H and F + H + I) compensated for the negative fertil-
izer effect on four species of this group (A. reptans, V. chamaedrys,
P. lanceolata, and A. capillaris); frequencies of these species were
either not significantly affected by the treatment combinations
(F + H and F + H + I) (V. chamaedrys, P. lanceolata, and A. capillaris)

Table 1
Mean plant frequency [%] of the most common dicotyledons (plant frequency > 1%) per plot and treatment in June 2012. A frequency of 100% was possible, when the species
was recorded in each sub-square of the vegetation assessments (in total 150 sub-squares per plot). Replicates per treatment = 8. Species are listed in descending order of
frequency. The last two columns show the results of the three-factorial PerAnova (=main effects and interaction effects). C = control, I = insecticide, F = fertilizer, H = herbicide.
SE = standard errors [%], P = perennial, A = annual.

Control Single treatments Combined treatments

Species C (± SE) I (± SE) F (± SE) H (± SE) H+I (± SE) F+I (± SE) F+H (± SE) F+H+I (± SE) Main effect Inter-action

1 Galium mollugo P 85.3 (3.4) 89.5 (2.8) 89.5 (2.2) 83.3 (6.4) 91.9* (2.4) 92.6** (3.1) 89.8 (2.3) 85.5 (4.5) / /

2 Glechoma hederacea P 71.8 (3.9) 65.3 (7.4) 50.3*** (8.7) 59.6* (4.9) 45.1*** (8.7) 65.0 (6.0) 34.3*** (5.4) 35.3*** (9.6) F**, H*** /

3 Ranunculus acris P 58.2 (4.1) 54.3 (3.6) 31.1*** (6.5) 45.4** (8.0) 52.3 (7.6) 23.3*** (3.5) 27.7*** (6.2) 33.2*** (5.5) F*** /

4 Lathyrus pratensis P 53.2 (7.6) 54.3 (5.0) 16.2*** (4.8) 27.8*** (5.5) 31.3*** (5.0) 17.4*** (5.6) 15.3*** (4.5) 12.5*** (4.3) F***, H*** FxH**
5 Vicia sepium P 49.6 (3.1) 45.8 (5.2) 28.0*** (5.6) 25.7*** (5.5) 30.7*** (3.8) 33.8*** (4.8) 11.2*** (3.5) 13.7*** (2.3) F***, H*** /
6 Ajuga reptans P 39.8 (4.8) 43.8 (5.4) 9.5*** (1.8) 38.8 (6.5) 32.3 (4.0) 8.3*** (1.7) 22.0** (6.3) 16.6*** (2.4) F*** FxH*
7 Rumex acetosa P 38.3 (4.5) 29.4 (4.4) 23.3*** (6.0) 21.5*** (4.0) 15.8*** (3.4) 23.4*** (5.1) 9.8*** (1.5) 10.8*** (1.5) F**, H*** /
8 Veronica chamaedrys P 33.3 (8.5) 38.0 (8.7) 11.3*** (3.6) 47.3 (7.1) 25.8 (6.0) 15.9** (6.4) 33.8 (8.7) 24.1 (7.9) F** /
9 Ranunculus repens P 25.6 (4.8) 20.5 (5.9) 15.8 (5.1) 6.0*** (3.3) 4.2*** (1.8) 12.1** (4.2) 1.3*** (0.6) 1.5*** (1.0) F*, H*** /
10 Plantago lanceolata P 16.4 (4.3) 13.3 (4.4) 5.3*** (1.8) 21.4 (6.0) 15.9 (5.6) 2.9*** (1.5) 11.2 (5.1) 10.8 (7.5) F** /
11 Vicia hirsuta A 14.8 (4.6) 8.4 (3.3) 1.3*** (0.5) 4.1*** (2.9) 4.0*** (2.9) 2.0*** (1.6) 0.4*** (0.2) 0.8*** (0.4) F***, H* /
12 Lotus corniculatus P 13.6 (4.7) 10.7 (4.4) 1.4*** (0.7) 13.1 (8.0) 17.1 (7.8) 3.3** (2.1) 3.9** (1.8) 8.3 (4.5) F** /
13 Rhinanthus alectorolophus A 11.0 (8.4) 6.3 (3.7) 0.2*** (0.1) 0.0*** (0.0) 0.1*** (0.1) 0.6*** (0.6) 0.0*** (0.0) 0.1*** (0.1) F*, H* FxH*
14 Calystegia sepium P 7.7 (4.7) 6.5 (3.9) 9.2 (4.9) 9.7 (4.8) 9.7 (6.4) 4.8 (2.3) 10.6 (4.8) 11.8 (5.6) / /
15 Stellaria graminea P 6.3 (3.6) 22.8*** (7.2) 17.3** (6.0) 1.1*** (0.6) 3.8 (1.3) 20.5*** (7.4) 0.3*** (0.2) 0.0*** (0.0) H***, I* /
16 Hypericum perforatum P 4.2 (2.4) 0.8 (0.7) 1.3 (1.3) 1.4 (1.4) 2.3 (1.5) 1.1 (0.7) 1.0 (0.8) 4.6 (2.6) / /
17 Leucanthemum vulgare P 3.4 (2.8) 1.1 (0.7) 0.8 (0.4) 6.7 (5.7) 2.8 (1.5) 0.2** (0.1) 0.1*** (0.1) 1.1 (1.0) / /
18 Achillea millefolium P 3.1 (1.6) 3.0 (0.8) 3.8 (1.8) 3.8 (2.0) 3.8 (1.9) 5.3 (2.0) 2.3 (1.2) 3.8 (1.9) / /
19 Heracleum sphondylium P 3.1 (1.2) 1.2 (0.6) 5.0 (2.8) 1.1 (0.4) 1.2 (0.4) 0.3** (0.2) 1.1 (0.6) 0.7* (0.3) / /

Grey background = plant frequency was significantly negatively affected by the treatment in comparison to the control, framed cells = plant frequency was significantly
positively affected by the treatment in comparison to the control. * p < 0.05, ** p < 0.01, *** p < 0.001, p values Bonferroni corrected, nested PerAnova.
64 J. Schmitz et al. / Agriculture, Ecosystems and Environment 193 (2014) 60–69

Table 2
Mean plant frequency [%] of the most common monocotyledons (plant frequency > 1%) per plot and treatment in June 2012. A frequency of 100% was possible, when the
species was recorded in each sub-square of the vegetation assessments (in total 150 sub-squares per plot). Replicates per treatment = 8. Species are listed in descending
order of frequency. The last two columns show the results of the three-factorial PerAnova (=main effects and interaction effects). C = control, I = insecticide, F = fertilizer,
H = herbicide. SE = standard errors [%], P = perennial.

Control Single treatments Combined treatments

Inter-
Species C (± SE) I (± SE) F (± SE) H (± SE) H+I (± SE) F+I (± SE) F+H (± SE) F+H+I (± SE) Main effect
action
20 Arrhenatherum elatius P 82.4 (4.8) 73.4 (5.5) 83.1 (5.9) 30.1*** (4.0) 30.8*** (4.4) 86.1 (3.3) 35.0*** (4.6) 34.9*** (5.3) H*** /
21 Dactylis glomerata P 56.1 (4.4) 48.1 (6.4) 68.3* (8.6) 53.9 (4.5) 60.3 (4.9) 59.8 (7.3) 72.3** (4.3) 72.0** (5.7) F** /
22 Agrostis capillaris P 26.8 (6.9) 26.2 (6.3) 7.6*** (2.4) 31.3 (7.2) 33.4 (4.4) 11.7** (4.9) 29.5 (5.5) 18.6 (4.7) F** /
23 Holcus lanatus P 26.6 (7.7) 19.8 (5.1) 16.5* (4.5) 26.3 (4.0) 19.3 (3.9) 13.2** (3.5) 9.5*** (2.5) 15.4** (4.0) F** /
24 Alopecurus pratensis P 6.8 (2.1) 2.3** (1.2) 10.8 (4.1) 1.3*** (1.0) 1.4*** (0.8) 3.3 (0.9) 0.5*** (0.2) 2.0** (0.8) H***, I* /
25 Festuca arundinacea P 5.4 (1.0) 2.7 (0.5) 1.9* (0.7) 11.7** (4.0) 12.3** (2.6) 2.7 (0.8) 5.2 (1.6) 9.4 (3.0) F*, H*** /
26 Anthoxanthum odoratum P 3.6 (1.5) 3.9 (2.0) 0.2*** (0.1) 0.3*** (0.2) 0.2*** (0.2) 0.0*** (0.0) 0.1*** (0.1) 0.0*** (0.0) F***, H*** FxH**

Grey background = plant frequency was significantly negatively affected by the treatment in comparison to the control, framed cells = plant frequency was significantly
positively affected by the treatment in comparison to the control. * p < 0.05, ** p < 0.01, *** p < 0.001, p values Bonferroni corrected, nested PerAnova.

or were less affected than in the fertilized plots (F) (A. reptans). A
significant herbicide–fertilizer interaction effect was detected for
one of these species: A. reptans (factorial PerAnova: F × H: p = 0.01)
(Table 1, last column).
neutralized the negative fertilizer effect in the F + H and F + H + I
treatments (Table 2).
D. glomerata was significantly positively affected by the separate
fertilizer treatment, whereas the herbicide treatment caused no
significant effect.

3.2.3. Response 3:
Six species (Galium mollugo, Calystegia sepium, Hypericum
perforatum, Leucanthemum vulgare, Achillea millefolium, and H.
sphondylium) were not significantly affected by the separate
fertilizer, herbicide, or insecticide treatment. We found no
herbicide–fertilizer interactions or main effects, despite the fact
that the one-way PerAnova showed significant effects for G.
mollugo (H + I and F + I), for L. vulgare (F + I and F + H) and for H.
sphondylium (F + I and F + H + I) (Table 1).
The remaining six species in Tables 1 and 2 (Ranunculus repens,
S. graminea, Arrhenatherum elatius, Dactylis glomerata, Alopecurus
pratensis, and Festuca arundinacea) were not assigned to response
groups, as they showed different reactions to the treatments. R.
repens, S. graminea, A. elatius, and A. pratensis were significantly
negatively affected by the herbicide, whereas they differed in their
responses to the insecticide and fertilizer treatment. For example,
the separate fertilizer and insecticide treatment increased the mean
frequency of S. graminea by factors of 2.75–3.62 (mean frequency
F: 17.3%; I: 22.8%) compared to the control (mean frequency 6.3%).
F. arundinacea also exhibited a unique response to the treat-
ments. The separate fertilizer treatment reduced its frequency,
whereas the separate herbicide treatment significantly increased
its frequency. Moreover, it seems that the positive herbicide effect
3.3. Community diversity
Species diversity (expressed as the mean Shannon index, Fig. 2)
was highest in the control and insecticide treated plots during
all 3 years of the study. The separate fertilizer and herbicide
treatments and their treatment combinations (F + H and F + H + I)
reduced species diversities compared to the control. These reduc-
tions were significant in the F + H + I, F + H and H + I treatments
during the first experimental season. In the third year, in all fertil-
izer and herbicide treatments a significantly lower species diversity
than in the control was observed (Fig. 2).
3.4. Community composition
Plant community compositions of the treated and control plots
were similar during the first and second years (Fig. 3A, B). The NMDS
diagram of the third year revealed three clearly distinct groups: one
group (1) consisted of the control and insecticide treated plots (C
and I), another group (2) consisted of the fertilizer treated plots (F
and F + I), and the last group (3) consisted of the herbicide treated
plots (H and H + I) and those receiving combinations of fertilizer
and herbicide (F + H and F + H + I) (Fig. 3C). We observed significant
differences between all three groups (one-way PerManova, p < 0.01

Table 3
Response groupsa of individual species to the treatments in June 2012. Separate treatments (F = fertilizer, H = herbicide, I = insecticide) are listed. Symbols within brackets
indicate whether the treatment caused a significant decrease (–) or no significant change (0) in the mean plant frequency in comparison to the control. Numbers next to the
species (within parentheses) correspond to the number designations of the species in Tables 1 and 2.

Response 1 Response 2 Response 3

F(–) H(–) I(0) F(–) H(0) I(0) F(0) H(0) I(0)

Dicotyledons G. hederacea (2) A. reptans (6) G. mollugo (1)

R. acris (3) V. chamaedrys (8) C. sepium (14)
L. pratensis (4) P. lanceolata (10) H. perforatum (16)
V. sepium (5) L. corniculatus (12) L. vulgare (17)
R. acetosa (7) A. millefolium (18)
V. hirsuta (11) H. sphondylium (19)
R. alectorolophus (13)

Monocotyledons A. odoratum (26) A. capillaris (22)

H. lanatus (23)
a
A response group was created when more than five individual species showed the same response to a treatment. Six species (R. repens (9), S. graminea (15), A. elatius (20),
D. glomerata (21), A. pratensis (24), and F. arundinacea (25)) could not be assigned to response groups.
 J. Schmitz et al. / Agriculture, Ecosystems and Environment 193 (2014) 60–69
Fig. 2. Species diversity per treatment and year (2010–2012) expressed as the mean
Shannon index [H = −SUM(Pi*log(Pi))]. The indices below the dotted lines (-----
) differ significantly from the control. Asterisks (*) indicate significance: p < 0.05
(one-way analysis of variance with permutations [PerAnova]; p values Bonferroni
corrected). C = control, F = fertilizer, H = herbicide, I = insecticide.
for all pairwise comparisons) in 2012, whereas treatments belong-
ing to one group (with one exception) did not differ significantly
from one another. The exception was the F + H and H treatment
(group 3), for which significant differences were detected (one-way
PerManova, p = 0.008, pairwise comparison).
The NMDS diagram with the weighted-average species scores
of June 2012 indicated that the species composition was affected
by the fertilizer and herbicide treatments (Fig. 3D). We found
no significant correlation between the species composition and
the insecticide treatment. Analysis of main effects (three facto-
rial PerManova) confirmed this finding and indicated a significant
main effect of fertilizer (p = 0.0001) and herbicide (p = 0.0001)
on the plant community composition in 2012. No significant
herbicide–fertilizer interactions were found at the community
level.
NMDS revealed that those species that were negatively affected
by the herbicide and fertilizer treatments (Response 1, Table 3)
were plotted near the control (Fig. 3D). In contrast, species
that were only negatively affected by the fertilizer treatment
(Response 2, Table 3) were plotted in the opposite direction of
fertilizer, i.e., between the herbicide and control, as their fre-
quencies were similar in the herbicide and control plots. Those
species that were not affected by the treatments (Response 3,
Table 3) displayed no discernible correlation with any factor
(Fig. 3D).
3.5. Biomass
The fresh weights of the biomass samples yielded similar results
each year. The mean fresh weights were highest in the plots receiv-
ing the fertilizer treatments (F and F + I), followed by those of the
control and insecticide treatments (Fig. 4). The lowest fresh weights
were found in the plots treated with the herbicide (H and H + I).
Analysis of main effects (three factorial PerAnova) showed a signif-
icant main effect of fertilizer and herbicide on the biomass every
year (p < 0.01). In addition, slightly significant herbicide–fertilizer
interactions were identified in June 2010 (factorial PerAnova, F × H,
p = 0.04) and June 2012 (factorial PerAnova, F × H, p = 0.02).
65
4. Discussion
4.1. Effects of fertilizer, herbicide, and insecticide on individual
plant species
Agrochemical effects on certain species were already observed
in the first year, but became stronger over time. As a result, 20 of
the 26 abundant species were significantly affected by the treat-
ments in the third year. Tables 1 and 2 show that 17 species were
significantly affected (15 negatively, 2 positively) by the separate
fertilizer treatment and that 13 species were significantly affected
(12 negatively, 1 positively) by the separate herbicide treatment.
The insecticide treatment affected 2 species significantly (1 posi-
tively, 1 negatively).
The fertilizer effect on several species is not surprising because
negative effects of fertilizers on plant species are well-documented
in the literature (e.g., Clark and Tilman, 2008; Kleijn et al., 2009;
Socher et al., 2013). Effects of eutrophication with low applica-
tion rates of fertilizer similar to the rates used in our study (25%
of the field rate) often develop slowly and reduce the population
size of plant species over time. Species losses or a decrease in cer-
tain species are often caused by competitive advantages of a few
nitrogen-tolerant species. Tall grasses such as D. glomerata (as in
our study) benefit from fertilization because they can increase in
abundance over smaller herbs (Socher et al., 2013). Small species
are usually replaced by faster-growing species (Jumpponen et al.,
2005) and by species with a high leaf canopy (Wilson, 1999). This
variation enhances effects of shading and, consequently, the com-
petition for light, which can further suppress the frequencies of
small and subordinate species (Hautier et al., 2009), such as G.
hederacea and A. reptans. Certain species with low nitrogen-values
(Ellenberg et al., 1992) were also negatively affected (e.g., R. alec-
torolophus (N-value = 3), V. hirsuta (N-value = 4)).
The herbicide affected the plant species in a different manner.
The herbicide treatment caused nearly a complete disappearance
of three species (R. alectorolophus, S. graminea, and A. odoratum)
in 2012 (Tables 1 and 2). Similar reductions were also observed
in the first and second years. The herbicide that was applied is
designed to control grasses and annual herbs. R. alectorolophus is an
annual, which may explain its high sensitivity, whereas S. graminea
is a perennial; thus, a generalization is difficult. These findings,
however, are of great concern because R. alectorolophus is on the
Red List “V” in Germany (V = vulnerable = species likely to become
endangered in the near future).
During the vegetation assessments, it was also observed that
the herbicide visibly affected the vegetation. Several leaves in the
herbicide treated plots were slightly yellow or brown. The affected
species were still recorded after the herbicide applications but, pre-
sumably, they were more vulnerable and sensitive to natural stress.
Their fitness and competitive ability were most likely reduced, and
the repeated herbicide applications intensified the effects, which
reduced the abundance of the most sensitive species over time.
This trend indicates that certain species were more sensitive to the
herbicide than others when growing in a natural plant community.
A few of the dicotyledons were also negatively affected by the
herbicide due to flower suppression. In the same field experiment,
significant reductions in the seed production of R. acris, L. praten-
sis, and V. sepium were observed in the herbicide-treated plots
(Schmitz et al., 2013, 2014). Such reductions in fruit sets during
one growing season, especially when combined with other stresses
(e.g., herbivores, weather conditions) are likely to be sufficient to
hamper recovery, and reproduction of most plants (Carpenter and
Boutin, 2010). These findings may explain why herbicide effects
often became apparent the first year after application. With recur-
ring exposure, the seed bank eventually will be depleted, which
will reduce population size (Roberts and Neilson, 1981; Ball, 1992).
66 J. Schmitz et al. / Agriculture, Ecosystems and Environment 193 (2014) 60–69

Fig. 3. A–C. Two-dimensional NMDS graph of the 64 plots in June of 2010 (A), 2011 (B), and 2012 (C). Plots are labeled to their treatments. Different symbols represent
different treatments and polygons enclose all plots of the same treatment. C = control, I = insecticide, F = fertilizer, H = herbicide (stress-values ranged from 0.20 to 0.25).
D. Two-dimensional NMDS graph of the weighted-average species scores (June 2012) (stress-value 0.21). Vectors significantly correlated with ordination (species composition)
are shown (fertilizer, r2 = 0.61, p = 0.001; herbicide, r2 = 0.79, p = 0.001; control, r2 = 0.19, p = 0.002). Numbers correspond to the designations assigned to the species in
Tables 1 and 2. The symbols represent species response groups summarized in Table 3: Response 1(䊉), Response 2 (o), Response 3 (), and species not assigned to response
groups ( ).

Other studies have also demonstrated that herbicides may reduce
seed sets when sprayed at the bud stage or shortly before the onset
of reproduction (Fletcher et al., 1996; Boutin et al., 2000; Kjaer et al.,
2006).
An additional herbicide effect was the inhibition of plant growth,
as indicated by the biomass samples. The biomass correlated well
with the vegetation height, which was also lowest in the herbi-
cide treated plots (Fig. 4, see Appendix A, Fig. S1 for vegetation
heights in June 2012). This inhibition increased the light inten-
sity and most likely was an advantage for less herbicide-sensitive
dicotyledons (A. reptans, V. chamaedrys, P. lanceolata, L. cornicula-
tus) and monocotyledons (D. glomerata, A. capillaris, H. lanatus, F.
arundinacea). Their frequencies were either not significantly nega-
tively affected or increased slightly (not significant) (V. chamaedrys,
P. lanceolata) or significantly (F. arundinacea) with the herbicide
treatment (Table 1).

Fig. 4. Mean fresh weights [g] of the biomass samples (1 m2 ) per plot and treatment in June 2010, 2011, and 2012. Replicates per treatment = 8. * Significantly different from
the control (one-way analysis of variance with permutations [PerAnova]; p values Bonferroni corrected; p < 0.01.). C = control, I = insecticide, F = fertilizer, H = herbicide.
 J. Schmitz et al. / Agriculture, Ecosystems and Environment 193 (2014) 60–69
The insecticide treatment increased the frequency of S. graminea
by a factor of 3.6 compared to the control. Similar increases were
observed during all 3 years. The insecticide is a pyrethroid, which
may have toxic or repellent effects for insects (Gist and Pless,
1985; Blair, 1991). For example, the micro-moth Coleophora stri-
atipennella uses the flowers of S. graminea for oviposition, and
the fruits/seeds are a food source for their larvae (Database of
Insects and their Food Plants, 2013). This treatment may have
had negative effects on insects (direct effect) and, in consequence,
positive effects on the plant (indirect effect) because certain her-
bivorous and seed eating insects decreased in number. In contrast,
the frequency of A. pratensis was significantly reduced by the
insecticide. Perhaps A. pratensis has mutualistic relationships with
specific arthropods. These possibilities, however, are only specula-
tions.
The results also indicated that there were no main effects on the
frequencies of six plant species (Table 1). Five of these species (C.
sepium, H. perforatum, L. vulgare, A. millefolium, and H. sphondylium)
were present in low frequencies (with high standard errors); thus,
it is difficult to develop firm conclusions regarding their sensitivity
to the agrochemicals. However, G. mollugo was the most abun-
dant species in the study site and seems to be not sensitive to the
fertilizer and herbicide products that were used in our study.
We used only one product per treatment (herbicide, and insecti-
cide). In general, species sensitivity varies greatly with the pesticide
used, and agrochemicals containing other active ingredients might
cause other effects. Therefore, it is difficult to extend our findings
to other agrochemicals. Moreover, we used formulated products in
our field experiment, and the ingredients in other formulations may
differ. These various ingredients may also influence species sensi-
tivity, which is largely unknown until now. Thus, it may be valuable
to investigate other agrochemicals and their effects on non-target
plants in future studies.
4.2. Interaction effects of fertilizer, herbicide, and insecticide on
individual plant species
We observed significant herbicide–fertilizer interaction effects
on four species (Tables 1 and 2). However, it seems that inter-
action effects between fertilizer and herbicide also occurred for
other species. For example, the herbicide neutralized the negative
fertilizer effects in the treatment combinations F + H and F + H + I
for V. chamaedrys, P. lanceolata, A. capillaris, and F. arundinacea.
Conversely, it appears likely that there were additive effects on
G. hederacea, V. sepium, and R. acetosa, as their frequencies were
negatively affected by the single fertilizer and herbicide treatment,
whereas their frequencies were more strongly reduced by the treat-
ment combinations of fertilizer and herbicide (F + H and F + H + I).
Thus, it is not possible to extrapolate from individual effects of her-
bicide and fertilizer to their combined effects on specific species.
There are two possible explanations: on the one hand, agrochem-
icals can interact with one another by neutralizing effects or by
causing synergistic or additive effects, as mentioned above. On
the other hand, the application of agrochemicals to natural plant
communities can (simultaneously) affect or influence the sensi-
tivity of certain species; thus, their competitiveness within the
community is altered. For example, Damgaard et al. (2011) inves-
tigated the combined effects of nitrogen and glyphosate on two
grasses (Festuca ovina and A. capillaris) and observed significant
positive interactions of glyphosate and nitrogen on the growth of A.
capillaris. The authors suggested that positive herbicide–fertilizer
interactions on this species were caused by altered plant competi-
tion (Damgaard et al., 2011).
Interaction effects in natural plant communities are complex
and depend on interaction effects between agrochemicals on
certain plant species (e.g., additive, synergistic effects) and on
67
interaction effects between species (e.g., changed competition,
shading effects). The relative intensity of species competition along
environmental gradients (e.g., agrochemicals) may also depend on
species density in the plant community (Damgaard and Fayolle,
2010). Thus, it is most likely that fertilizer–herbicide interaction
effects differ with the type of vegetation. Interaction effects of agro-
chemicals may also vary with the applied fertilizer and herbicide
products due to different modes of action.
4.3. Effects of fertilizer, herbicide, and insecticide on the plant
community composition
Before the experiment began, the study site was not con-
taminated with fertilizer and/or pesticides and thus, the plant
community may be regarded as a habitat free of any influence of
agrochemicals. Plant communities of unaffected habitats are rela-
tively stable in their composition and change only slowly following
low fertilizer or pesticide application rates. These effects take time
to appear, as reflected by the results of the NMDS analysis. Although
the plant frequencies of a few species were already affected during
the first and second years, these effects were most likely too weak
to cause clear separations of the plant communities in the differ-
ent treatments (Fig. 3A, B). With each year of application, effects on
the frequencies of individual species became stronger, and after 3
years the composition of the plant community was altered by the
herbicide and fertilizer treatments (Fig. 3C, D). The NMDS diagram
of the 64 plots based on the June 2012 data shows a separation of
the fertilizer-treated plots (F and F + I) from all other treatments
(Fig. 3C). This result illustrates the severe effect of fertilizer on the
frequencies of several species. Herbicide treatments (H and H + I)
and treatment combinations of fertilizer and herbicide (F + H and
F + H + I) are plotted closer to each other in the NMDS diagram due
to the above-mentioned herbicide–fertilizer interaction effects. It
is likely that a stronger separation of the treatment combinations
(F + H and F + H + I) and the herbicide-treated plots (H and H + I)
would occur in the course of time, based on the fact that a signifi-
cant difference between the F + H and H treatments was observed
in June 2012.
The number of species was slightly reduced in the treatment
combinations of fertilizer and herbicide in June 2012 (Fig. 1).
Although the separate fertilizer and herbicide treatments did not
significantly affect the mean number of species, we observed sig-
nificantly lower species diversities in all fertilizer and herbicide
treatments compared to the control in the third year (Fig. 2). It
may be assumed that these effects would become more pronounced
with each year of agrochemical applications until the most robust
and least susceptible species dominate the plant community.
5. Conclusion
This study revealed that fertilizer and herbicide misplacements
in field margins are major factors that affect field margin plant com-
munities. Previous studies have often focused only on the effects
of herbicide drift on the vegetation in off-field habitats, and the
effects of fertilizer were not considered. Our results demonstrate
that both agrochemicals – herbicide and fertilizer – can reduce
mean frequencies of several species, although these agrochemicals
affect different species in different ways. In our study, fertilizer
directly promoted the growth and spread of species with a high
nutrient uptake, whereas it indirectly reduced the growth of small
species. In contrast, the herbicide caused predominantly sublethal
effects (phytotoxic and reproduction effects) and nearly a complete
disappearance of a few species.
Reproduction effects require time to be measurable in plant
frequency assessments; nonetheless, these effects are expected to
68 J. Schmitz et al. / Agriculture, Ecosystems and Environment 193 (2014) 60–69
reduce population size of plant species in the long term. Thus, inves-
tigations that only focus on short-term herbicide effects on mean
plant frequencies may underestimate the full herbicide effect. In
our field study, negative effects on plant frequencies and plant
reproduction, contributed together to the alterations of the plant
communities. Moreover, the application of the insecticide signif-
icantly affected the frequencies of two species, which may have
resulted from effects on plant–insect interactions, although this
assumption requires more specific investigations.
An additional key finding was that treatment combinations of
fertilizer and herbicide caused interaction effects (e.g., additive
effects) and that these effects could not be extrapolated from the
individual fertilizer or individual herbicide effect.
Treatment effects (individual and combined) on most species
gradually became apparent. Thus, low fertilizer and herbicide rates
may not significantly affect plant community compositions in field
margins in the short-term (within 1 or 2 years), but long-term
misplacements of agrochemicals may cause significant changes
in species composition and reduce species diversity significantly,
as was observed in our study. We believe that continuous annual
applications of agrochemicals on the study site would cause further
plant community shifts and would likely lead to the disappear-
ance of certain affected plants. Small and subordinate species with
high herbicide and fertilizer sensitivity will most likely become
less abundant or completely disappear. The most robust and least
susceptible species (predominantly tall grasses and a few species
of dicotyledons) may dominate the plant community after several
years.
Narrow field margins in Germany are most likely similarly
affected, as simulated in our study. As a result, the vegetation in
these margins has already been degraded due to agrochemical
inputs over the last five to six decades. Although herbicide risk
assessment (RA) aims to protect non-target plants in off-field habi-
tats (field margins) from adverse effects, reproduction effects are
not considered so far. The RA of pesticides is also carried out for
only one specific compound, and therefore, influences of combined
effects of pesticides are not taken into account. Therefore, it seems
that the RA currently performed provides insufficient protection for
non-target plants and their habitats. Furthermore, there are no dis-
tance requirements for fertilizer applications next to field margins,
which could prevent fertilizer misplacements.
Adaptations of the current RA and development of general risk
mitigation measures (e.g., in-field buffers) for the application of
herbicides and fertilizers are urgently required to conserve natural
plant communities in field margins.
Acknowledgments
This study was supported by a scholarship from the Univer-
sity Koblenz-Landau, financed by the Lotto-Stiftung Rhineland-
Palatinate (Germany), to Juliane Schmitz. We would like to express
our gratitude to: the Schnörringer familiy and Bruno Kreiselmeyer
for their friendly cooperation, to Lars Hartmann for his technical
support and assistance in the field and to all the students partici-
pating in this project. We would also like to thank Martin Entling
for statistical advice, Horst Tremp for sharing his superb botan-
ical knowledge with us and two anonymous reviewers for their
thorough and constructive comments.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.agee.2014.04.025.
References
Anderson, M.J., Gorley, R.N., Clarke, K.R. (Eds.), 2008. Permanova+ for Primer: Guide
to Software and Statistical Methods. Primer-E Ltd., Polymouth, UK.
Asteraki, E.J., Hart, B.J., Ings, T.C., Manley, W.J., 2004. Factors influencing the plant
and invertebrate diversity of arable field margins. Agr. Ecosyst. Environ. 102,
219–231.
Ball, D.A., 1992. Weed seedbank response to tillage, herbicides, and crop rotation
sequence. Weed Sci. 40, 654–659.
Benton, T.G., Vickery, J.A., Wilson, J.D., 2003. Farmland biodiversity: is habitat het-
erogeneity the key? Trends Ecol. Evol. 18, 182–188.
Berendse, F., Chamberlain, D., Kleijn, D., Schekkerman, H., 2004. Declining biodi-
versity in agricultural landscapes and the effectiveness of agri-environment
schemes. Ambio 33, 499–502.
Blair, B.W., 1991. The concentration of lamdacyhalothrin and its effect on the cut-
worm Agrotis segetum (Lepidoptera: Noctuidae). Bull. Entomol. Res. 81, 143–145.
Boutin, C., Lee, H.B., Peart, E.T., Batchelor, P.S., Maguire, R.J., 2000. Effects of the
sulfonylurea herbicide metsulfuron methyl on growth and reproduction of five
wetland and terrestrial plant species. Environ. Toxicol. Chem. 19, 2532–2541.
BVL, 2013. “Kodeliste für Kennzeichnungstexte und sonstige Auflagen zum
Pflanzenschutzmittelverzeichnis”, Last accessed January 2013, see http://www.
bvl.bund.de/DE/04 Pflanzenschutzmittel/01 Aufgaben/02 ZulassungPSM/01
ZugelPSM/01 OnlineDatenbank/psm onlineDB node.html
Carpenter, D., Boutin, C., 2010. Sublethal effects of the herbicide glufosinate ammo-
nium on crops and wild plants: short-term effects compared to vegetative
recovery and plant reproduction. Ecotoxicology 19, 1322–1336.
Clark, C.M., Tilman, D., 2008. Loss of plant species after chronic low-level nitrogen
deposition to prairie grasslands. Nature 451, 712–715.
Damgaard, C., Fayolle, A., 2010. Measuring the importance of competition: a new
formulation of the problem. J. Ecol. 98, 1–6.
Damgaard, C., Strandberg, B., Mathiassen, S.K., Kudsk, P., 2011. The combined effect
of nitrogen and glyphosate on the competitive growth, survival and estab-
lishment of Festuca ovina and Agrostis capillaris. Agric. Ecosyst. Environ. 142,
374–381.
Database of Insects and their Food Plants, 2013. Last assessed October 2013, see
http://www.brc.ac.uk/dbif/hostsresults.aspx?hostid=5408
de Snoo, G.R., van der Poll, R.J., 1999. Effect of herbicide drift on adjacent boundary
vegetation. Agric. Ecosyst. Environ. 73, 1–6.
Ellenberg, H., Weber, H.E., Düll, R., Wirth, V., Werner, W., Paulißen, D., 1992. Zeiger-
werte der Pflanzen in Mitteleuropa. Verlag Erich Goltze KG, Göttingen.
Elzinga, C.L., Salzer, D.W., Willoughby, J.W. (Eds.), 1998. Measuring and Monitor-
ing Plant Populations. Bureau of Land Management, National Business Center,
Denver, Colorado.
Firbank, L.G., Petit, S., Smart, S., Blain, A., Fuller, R.J., 2008. Assessing the impacts of
agricultural intensification on biodiversity: a British perspective. Philos. Trans.
R. Soc. B – Biol. Sci. 363, 777–787.
Fletcher, J.S., Pfleeger, T.G., Ratsch, H.C., Hayes, R., 1996. Potential impact of low
levels of chlorsulfuron and other herbicides on growth and yield of nontarget
plants. Environ. Toxicol. Chem. 15, 1189–1196.
Freier, B., Pallutt, B., Jahn, M., Sellmann, J., Gutsche, V., Zornbach, W., 2008. Netz Ver-
gleichsbetriebe Pflanzenschutz: Jahresbericht 2007. Julius Kühn-Institut (JKI),
Bundesforschungsinstitut für Kulturpflanzen.
Gist, G.L., Pless, C.D., 1985. Ovicidal activity and ovipositional repellent properties of
synthetic pyrethroids to the fall armyworm, Spodoptera frugiperda. Fla. Entomol.
68, 462–466.
Gove, B., Power, S.A., Buckley, G.P., Ghazoul, J., 2007. Effects of herbicide spray drift
and fertilizer overspread on selected species of woodland ground flora: compar-
ison between short-term and long-term impact assessments and field surveys.
J. Appl. Ecol. 44, 374–384.
Hahn, M., Lenhardt, P., Brühl, C.A., 2014. Characterization of field margins in inten-
sified agro-ecosystems – why narrow margins should matter in terrestrial
pesticide risk assessment and management. Integr. Environ. Assess. Manage.,
http://dx.doi.org/10.1002/ieam.1535.
Hautier, Y., Niklaus, P.A., Hector, A., 2009. Competition for light causes plant biodi-
versity loss after eutrophication. Science 324, 636–638.
Jumpponen, A., Mulder, C.P.H., Huss-Danell, K., Högberg, P., 2005. Winners and losers
in herbaceous plant communities: insights from folier carbon isotope composi-
tion in monocultures and mixtures. J. Ecol. 93, 1136–1147.
Kindt, R., Coe, R., 2005. Tree diversity analysis. In: A Manual and Software for
Common Statistical Methods for Ecological and Biodiversity Studies. World
Agroforestry Centre (ICRAF), Nairobi.
Kjaer, C., Strandberg, M., Erlandsen, M., 2006. Metsulfuron spray drift reduces fruit
yield of hawthorn (Crataegus monogyna L.). Sci. Total Environ. 356, 228–234.
Kleijn, D., Kohler, F., Baldi, A., Batary, P., Concepcion, E.D., Clough, Y., Diaz, M., Gabriel,
D., Holzschuh, A., Knop, E., Kovacs, A., Marshall, E.J.P., Tscharntke, T., Verhulst, J.,
2009. On the relationship between farmland biodiversity and land-use intensity
in Europe. Proc. R. Soc. B-Biol. Sci. 276, 903–909.
Kleijn, D., Snoeijing, G.I.J., 1997. Field boundary vegetation and the effects of agro-
chemical drift: botanical change caused by low levels of herbicide and fertilizer.
J. Appl. Ecol. 34, 1413–1425.
Kleijn, D., Verbeek, M., 2000. Factors affecting the apecies composition of arable field
boundary vegetation. J. Appl. Ecol. 37, 256–266.
Kühne, S., Enzian, S., Jüttersonke, B., Freier, B., Forster, R., Rothert, H., 2000. Beschaf-
fenheit und Funktion von Saumstrukturen in der Bundesrepublik Deutschland
und ihre Berücksichtgung im Zulassungsverfahren im Hinblick auf die Schonung
 J. Schmitz et al. / Agriculture, Ecosystems and Environment 193 (2014) 60–69
von Nichtzielarthropoden. Mitteiltung der Biologische Bundesanstalt für Land-
und Forstwirtschaft, Berlin-Dahlem, Germany.
Kühne, S., Freier, B., 2001. Saumbiotope in Deutschland - ihre historische
Entwicklung, Beschaffenheit und Typisierung. Biozönosen in Saumbiotopen
in landwirtschaftlichen Einflussbereichen: Beeinflussung durch Pflanzen-
schutzmitteleinträge? Mitteiling der Biologsichen Bundesanstalt für Land- und
Forstwirtschaft, Berlin-Dahlem, Germany.
Marrs, R.H., Frost, A.J., 1997. A microcosm approach to the detection of the effects
of herbicide spray drift in plant communities. J. Environ. Manage. 50, 369–388.
Marshall, E.J.R., Moonen, A.C., 2002. Field margins in northern Europe: their func-
tions and interactions with agriculture. Agric. Ecosyst. Environ. 89, 5–21.
Nentwig, W., 2000. Die Bedeutung von streifenförmigen Strukturen in der Kultur-
landschaft. In: Nentwig, W. (Ed.), Streifenförmige ökologische Ausgleichsflächen
in der Kulturlandschaft: Ackerkrautstreifen, Buntbrache, Feldränder. Verlag
Agrarökologie, Bern, Hannover, pp. 11–40.
Oksanen, J., 2013. Multivariate Analysis of Ecological Communities in R: Vegan Tuto-
rial. R Foundation for Statistical Computing, pp. 1–43.
Perry, N.H., Chaney, K., Wilcox, A., Boatman, N.D., 1996. The effect of fertiliser and
herbicide application on herbaceous field margin communities. In: Vegetation
Management in Forestry, Amenity and Conservation Areas: Managing for Mul-
tiple Objectives; Aspects of Applied Biology. Association of Applied Biologists,
Wellesbourne, pp. 339–344.
Rew, L.J., Froud-Williams, R.J., Boatman, N.D., 1995. The effect of nitrogen, plant
density and competition between Bromus sterilis and three perennial grasses:
the implications for boundary strip management. Weed Res. 35, 363–368.
Roberts, H.A., Neilson, J.E., 1981. Changes in the soil seed bank of four long-term
crop/herbicide experiments. J. Appl. Ecol. 18, 661–668.
69
Schmitz, J., Schäfer, K., Brühl, C.A., 2013. Agrochemicals in field margins – Assessing
the impacts of herbicides, insecticides, and fertilizer on the common Buttercup
(Ranunculus acris). Environ. Toxicol. Chem. 32, 1124–1131.
Schmitz, J., Schäfer, K., Brühl, C.A., 2014. Agrochemicals in field margins –
Field evaluation of plant reproduction effects. Agric. Ecosyst. Environ. 189,
82–91.
Socher, S.A., Prati, D., Boch, S., Müller, J., Baumbach, H., Gockel, S., Hemp, A., Schöning,
I., Wells, K., Buscot, F., Kalko, E.K.V., Linsenmair, K.E., Schulze, E.-D., Weisser,
W.W., Fischer, M., 2013. Interacting effects of fertilization, mowing and grazing
on plant species diversity of 1500 grasslands in Germany differ between regions.
Basic Appl. Ecol. 14, 126–136.
Strandberg, B., Mathiassen, S.K., Bruus, M., Kjær, C., Damgaard, C., Andersen, H.V.,
Bossi, R., Løfstrøm1, P., Larsen, S.E., Bak, J., Kudsk, P., 2012. Effects of herbicides
on non-target plants: How do effects in standard plant test relate to effects in
natural habitats? Danish Environmental Protection Agency.
Tscharntke, T., Klein, A.M., Kruess, A., Steffan-Dewenter, I., Thies, C., 2005. Landscape
perspectives on agricultural intensification and biodiversity – ecosystem service
management. Ecol. Lett. 8, 857–874.
Tsiouris, S., Marshall, E.J.P., 1998. Observations on patterns of granular fertiliser
deposition beside hedges and its likely effects on the botanical composition of
field margins. Ann. Appl. Biol. 132, 115–127.
Wilson, P.J., 1999. The effect of nitrogen on populations of rare arable plants in
Britain. In: Boatman, N.D., Davis, B.N.K., Chaney, K., Feber, R., De Snoo, G.R.,
Sparks, T.H. (Eds.), Field Margins and Buffer Zones: Ecology, Management and
Policy. The Association of Applied Biologists, c/o Horticulture Research Interna-
tional, Wellesbourne, Warwick, UK, Wellesbourne, pp. 93–100.